Stereo Tool

You don't realize what good sound is unless you hear it first. For all this while each time Hans said that current implementation of ND has artifacts, I was like "it already sound great! It can't sound better.."
Anyways, I tried out the first beta of the newer implementation, and impressed by the clarity and re-infusion of dynamics.

Somehow, in the quick settings that I made the Peak detection mode sounded better - could be subjective. Will come back to it later.

Hans,
For now, I have two requests, should you decide to include them:
1. A toggle for before/after AGC as it was in previous implementation
2. A drive in the range of - 40 dB -> + 40 dB in ND

Please let me know if it's possible....

_________________
visit website

@Phoenix: The old ND responded to a specific target level, the new one does not. It does change the level though, that's why I placed it before the AGC. Placing it after it doesn't change the sound much (probably only at moments of big volume jumps), but will make the volume differences bigger - that's why I didn't include the before/after thing.

What do you mean with this Drive thing? (So: What should it do?)

By 'Drive' I mean a slider that could provide a fixed increment/decrement over the incoming level like you have one in MB and some others.

Also I have a valid reason for the toggle between Before/After AGC switch. When you listen to new ND in isolation, it sounds so enhanced. But soon after AGC is turned on, it becomes flattened.
I am pretty sure we could have substantial benefits in making ND act after AGC. Let me know your thoughts!

_________________
visit website

Lol, no. I could for the declipper but this ND thing is mainly guess work. It doesn't seem to sound bad though (since you have much more experience with things like this than I do, I should ask you: do you hear bad things in it?)

I agree that it takes time to 'learn' to listen to new filters - regardless whether it's an AGC, compressor, clipper, it takes time before you start hearing the bad things they do. And that's probably the same with Natural Dynamics - which is actually even worse because we're already used to hearing all the other things I mentioned, but not this.

I'm going to take another look at the first part of the filter (the part without Punch Protection). I'm pretty sure that Punch Protection does a good job, I actually looked at many waveforms to determine how it should work. But that first part was thrown together way too fast, I actually had a 'brilliant' idea of how to make it, then started programming and ran into difficulties. So then I just made something else to see if it would work, and that's still there - and in some cases it behaves badly (big volume changes).


I'm not going to tell you how I made this thing, I can tell you one thing though: I read a bit about dynamics expanders (opposite of compressors), and that is not at all what ND is. (Which explains why there are no things in it like attack, release etc.).

There should be no difference! Except level.

The input level is completely irrelevant for how ND works. The only thing that I would like to add is a global level to control how strongly it works.
The AGC doesn't remove the ND effect, it just adjusts the level!

If you've followed my "saga" with this computer, I've been due for a new system since about 2009. The difficulty has been the total lack of income (unemployment). That has recently changed, and I could take out a loan, but right now I'm hoping to coast along until I've saved enough.

As for MS direction, agreed. There really needs to be two product lines, or at least the ability for the end-user to change between a touch/tablet interface and a desktop/workstation interface. I don't have a tablet or a touch screen, so even with Windows Blue (8.1), there just isn't enough improvement in usability for me to warrant using it.

Need to spend substantial time this weekend with the new implementation. I didn't alter other parameters, except for switching from Old ND to new one. Actually I don't need to if what you and Bojcha are saying is true - somehow I am finding it difficult to perceive.

If threshold at which a band begins to expand the levels is absolute then probably all I need is to alter the detection levels or so...

Moreover, the UI in VST plugin for ND is so cramped up that I was having a hard time trying to adjust the levels.
Hope I come across an ND configuration that suits my AGC settings little better.

_________________
visit website

Ok so I did make a mistake in this version... And Punch is kinda broken. CPU load should be about the same as without the mistake so this version might still be usable to determine the CPU load improvement. Next version will have this improvement and sound good again.

Winamp version 725-002A: http://www.stereotool.com/download/dsp_ ... 5-002A.exe

Sound is nearly(*) identical to 725-002, but performance should now really be better - I've optimized the Punch part of the code (which I think is already ok).

For those of you interested in the changes: I had a number of difficult calculations that I'm now doing once in advance instead of repeating them continuously, and I had some if-statements in loops that I moved outside. And added some vectorizing.

I've checked if it was supporting multicore already, and it was - so I haven't changed that.


(*) I had to change some values such as look-ahead times to multiples of 4 samples to be able to optimize the code further. This should have no noticeable impact on the sound.